That NIH study assumes no injury would have occured in homes with a gun if no gun had been present. That doesn't seem too rigorous. US Bureau of Justice Statistics estimates Americans successfully use a gun in self defense ~115,000 times/year. The study acknowledges 13 such instances. Thirteen.

Yeah, I'd call it a fail if I have to shoot someone in a defensive gun use, I'd far rather deter them from further acts. Shooting is better than me getting hurt, but it's not optimal.